Estimating uncertainty in end-of-life costs and embodied carbon of construction projects

The building and construction sector is one of the most environmentally consequential industries on the planet. According to the United Nations Environment Programme, it accounts for 37% of global carbon emissions and 34% of worldwide energy use, while also generating roughly 30% of the planet’s solid waste. The scale of the problem is staggering: the United States alone produces 600 million tons of construction and demolition (C&D) waste every year—double the amount of its municipal solid waste. Other regions report even more troubling figures. China generates approximately 2,300 million tons of C&D waste annually, and the European Union produces 834 million tons. At the same time, the construction industry is an economic juggernaut, with global expenditures of roughly $10 trillion per year, making cost performance and benchmarking central to every project decision.

Against this backdrop, the circular economy has emerged as one of the most promising sustainability strategies in construction. Rather than demolishing buildings and sending the debris to landfills, circular approaches emphasize deconstruction—carefully disassembling structures so that beams, slabs, blocks, and other components can be salvaged, reused, remanufactured, or recycled in new projects. In some cases, this can simultaneously reduce embodied carbon (the emissions associated with producing construction materials), divert waste from landfills, and lower project costs. Yet industry practitioners remain hesitant, and one reason stands out above the rest: uncertainty. Cost is consistently identified by professionals as one of the main impediments to adopting carbon-reduction measures, and salvaged materials introduce a host of unknowns that virgin products simply do not have.

That uncertainty has now been quantified. A new study published in the open-access journal Cleaner Engineering and Technology by Alberto E. Pozzer, Nikiforos Repousis, Fernanda Leite, and Christopher Rausch presents the first comprehensive framework for measuring the uncertainty in both embodied carbon and cost assessments for reclaimed construction products. The research addresses a glaring gap in the scientific literature. While previous studies have applied uncertainty analysis to the manufacturing of construction materials, to whole-building product stages, and even to full project lifecycles, none had extended that analysis to the end-of-life and “beyond-life” stages—where materials from one project are salvaged and given a second life in another.

The central hypothesis of the research was straightforward but consequential: reused products likely have a lower mean embodied carbon and cost compared with virgin materials, but they inherently carry higher variability. If that variability goes unmeasured, procurement decisions based on average values alone could be misleading, or even wrong. As the authors put it, quantifying this spread is essential to ensure “meaningful and transparent procurement decision-making.” The study also posed a follow-up question with real-world implications: does a higher level of uncertainty itself influence whether decision-makers choose circularity in the first place?

To model that uncertainty, the researchers turned to Monte Carlo simulation, a computational technique that has become the workhorse of uncertainty analysis in lifecycle assessment (LCA). Rather than producing a single deterministic estimate of embodied carbon or cost, Monte Carlo simulation treats key input parameters as random variables, each characterized by a probability distribution. The model then generates random samples from these distributions—10,000 iterations in this study, a figure consistent with prior research showing that results converge at that scale—yielding thousands of possible outcomes. From these, the team derived means, standard deviations, and full probability density functions describing the range of plausible results.

Identifying which variables to model was itself a substantial undertaking. Drawing on a scoping review of prior literature, the researchers distilled the uncertainty sources specific to salvaged materials into measurable variables across four scopes: project end-of-life, transportation, reclaimed material properties, and beyond-life pathways. At the project end-of-life stage, the uncertainty of the deconstruction schedule and the emission factors of the equipment used become critical, since deconstruction demands more detailed planning than conventional demolition, and studies have flagged risks such as inaccurate labor estimates, insufficient skilled workers, and inadequate heavy equipment. Transportation introduces variability in distances and vehicle emission factors. The reclaimed materials themselves bring perhaps the thorniest uncertainties: their quantity, their properties—which the team modeled as a binary pass/fail against required conditions—and their durability, expressed as an uncertain remaining service life. Finally, the beyond-life stage adds scenario alternatives (reuse, recycling, remanufacturing) and process emission factors.

The study adopted a pragmatic approach to characterizing these variables statistically. Using @RISK, a Microsoft Excel–based risk analysis tool, the team identified best-fitting probability distributions wherever sufficient data existed. Where data was sparse, they applied simpler distributions: uniform distributions when only minimum and maximum values were known, and triangular distributions when an expected value, minimum, and maximum were available. This flexibility matters, because previous work has shown that when Monte Carlo simulations exceed 10,000 iterations, results tend to converge regardless of whether inputs are characterized as normal, uniform, or lognormal—meaning the choice of distribution matters less than capturing the plausible range. For virgin materials, the team used the same uncertainty factors established in earlier studies of the product, transportation, and construction stages.

The lifecycle accounting followed ISO 14040 and ISO 14044 standards, organized into four stages: the product stage (A1–A3), covering raw material extraction, transportation, and manufacturing for virgin products; the use stage (B1–B5), covering maintenance, repair, replacement, and refurbishment for both virgin and salvaged materials; the end-of-life stage (C), covering deconstruction and transportation of reclaimed materials; and stage D, covering reuse and recycling. The embodied carbon equations summed material quantities multiplied by emission factors, equipment fuel consumption multiplied by fuel emission factors, and transportation distances multiplied by vehicle emission factors, with additional equations accounting for the project lifespan and the service life of reused components. The cost model mirrored the carbon model exactly, with emission factors replaced by unit costs—a structure aligned with traditional quantity-times-unit-price cost estimation. This parallel structure is significant, because cost estimates themselves suffer from uncertainty in cost data and gaps in project scope definition, yet cost uncertainty is rarely analyzed alongside carbon uncertainty.

To demonstrate the framework, the researchers built a case study anchored in real project data. The end-of-life data—schedule, equipment, transport logistics—came from a selective demolition project previously documented by Repousis, while emission factors were drawn from the OneClickLCA Building LCA software, an industry-standard tool containing a comprehensive repository of lifecycle inventory data and environmental product declarations. Equipment fuel consumption rates came from technical sheets and external sources. The beyond-life scenario was constructed by defining a hypothetical new project based on a masterplan, drawings, and specifications, allowing the team to simulate what would happen if the salvaged materials were incorporated into future construction.

The simulation results were then compared probabilistically. For each material alternative—new, reused, and recycled—the team computed the probability of not exceeding a reference embodied carbon and a reference cost established as project goals. This is the framework’s key innovation: instead of declaring one option “better” based on mean values, it tells decision-makers how confident they can be that a given option will actually meet their carbon and cost targets. A reused material with a lower average embodied carbon but wide variability may carry a lower probability of hitting a strict carbon target than a virgin material with a modest but tightly clustered footprint—and the framework makes that trade-off visible.

The significance of this work extends well beyond a single case study. The literature on concrete reuse illustrates why: reported carbon savings from reusing concrete elements have ranged wildly, from 40% to 82% across different studies, and some analyses of recycled aggregate concrete have even projected emissions increases rather than reductions. Case-specific studies—such as assessments of reusing concrete blocks for a pedestrian bridge, or repurposing beams, floors, columns, and hollow-core slabs—have demonstrated potential but never incorporated uncertainty analysis. Reviews of circular economy research have repeatedly flagged the lack of knowledge and data on the quality of recovered and recycled materials as a barrier to implementation, and have noted that the accuracy of LCAs for upcycling demolition waste is constrained by data uncertainty. Until now, however, no study had actually quantified that uncertainty.

The framework is also deliberately replicable. The authors emphasize that uncertainty sources in LCA are affected by geographical, temporal, and technological representativeness—emission factors vary by region, and practices for demolition, deconstruction, and transportation differ from one location to another. The value of the approach, therefore, lies not only in the results of this particular case but in providing a method that practitioners can adapt to their own projects, plugging in locally relevant input variables. There are limits, too: the study deliberately excludes supply chain and market dynamics—demand fluctuations, supplier availability—although the authors acknowledge these external factors as important candidates for future research.

What emerges is a practical answer to a question that has long haunted sustainable construction. Reused materials do tend to offer lower average embodied carbon and cost, but that advantage comes wrapped in variability that must be measured, not ignored. By quantifying that variability, the new framework gives architects, engineers, and contractors a way to weigh circular options with their eyes open—transforming salvage from a leap of faith into a calculated, transparent decision.